Skip to content

feat(recovery): model vanished runner job recovery - #204

Merged
rldyourmnd merged 10 commits into
mainfrom
feat/vanished-runner-recovery-policy
Aug 26, 2026
Merged

feat(recovery): model vanished runner job recovery#204
rldyourmnd merged 10 commits into
mainfrom
feat/vanished-runner-recovery-policy

Conversation

@rldyourmnd

@rldyourmnd rldyourmnd commented Aug 26, 2026

Copy link
Copy Markdown
Contributor

Implements the autonomous recovery engine for #201.

The branch now includes:

  • strict typed policy and crash-reconstructable decision state;
  • fsync-backed compare-and-swap transaction store;
  • exact GitHub force-cancel and full-run rerun client;
  • gha-vanished-job-recovery plan/apply CLI;
  • bounded command observation for autonomous batch reconciliation;
  • per-repository bounded token-command resolution;
  • gha-garm-app-token, which mints short-lived installation tokens from an existing encrypted GARM GitHub App envelope without persisting private keys or tokens;
  • duplicate-observation rejection and bounded durable history;
  • unit and race coverage for policy, state, store, controller, GitHub API, observer and credential minting.

An incremented authoritative run_attempt suppresses duplicate reruns after a process crash, while completed records suppress replay within bounded history. Private runtime wiring is staged in NDDev-it-com/github-device-sync-estate#265; its live read-only qualification minted tokens and read organization runners for all three installed Apps without printing secrets.

Verification: make verify (vet, full race suite, trimpath build).

@rldyourmnd

Copy link
Copy Markdown
Contributor Author

Retriggering the exact verified head after GitHub Actions began recovering from the major outage; no code or commit identity changed.

@rldyourmnd rldyourmnd closed this Aug 26, 2026
@rldyourmnd rldyourmnd reopened this Aug 26, 2026
@rldyourmnd
rldyourmnd force-pushed the feat/vanished-runner-recovery-policy branch from 8802a76 to a043004 Compare August 26, 2026 16:29
@rldyourmnd

Copy link
Copy Markdown
Contributor Author

Retriggering the exact fully verified head after GitHub Actions resumed accepting traffic. Workflow-dispatch run 32989549278 passed Test, Race, Static/build, both reproducible derivatives and Gate on this exact commit.

@rldyourmnd rldyourmnd closed this Aug 26, 2026
@rldyourmnd rldyourmnd reopened this Aug 26, 2026
@rldyourmnd

Copy link
Copy Markdown
Contributor Author

Retriggering the exact current head after obsolete delayed runs from pre-rebase commits terminalized. Exact-head workflow-dispatch Gate 32989549278 is green.

@rldyourmnd rldyourmnd closed this Aug 26, 2026
@rldyourmnd rldyourmnd reopened this Aug 26, 2026
@rldyourmnd
rldyourmnd enabled auto-merge August 26, 2026 16:50
@rldyourmnd
rldyourmnd merged commit d87e995 into main Aug 26, 2026
17 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant